Citations

Introduction: Carotid artery intraplaque hemorrhage (IPH) is increasingly recognized as a predictive marker of cerebrovascular events. Despite widespread utilization of contrast-enhanced CT angiography (CTA) for the evaluation of carotid atherosclerosis, diagnosis of IPH on CTA is challenging, limiting risk stratification capabilities. We evaluated convolutional neural networks (CNN) for detection of histology-proven IPH in carotid CTA. Methods: The study included carotid endarterectomy patients from our center who had preoperative CTA and available plaque specimens from the Mayo Clinic Carotid Artery Atherosclerosis Trial (MCAT) Biobank. A validated segmentation model localized common and internal carotid arteries (CCA/ICA). CCA bifurcation served as landmark after a large language model, Llama3.0-70B, verified culprit plaque proximity to the bifurcation from CTA reports. A 3×3×4cm box around ipsilateral CCA bifurcation was sampled from CTA volumes and images were labeled IPH by histology study of carotid specimen (Fig 1). Given limited cases with concurrent imaging and histology, we employed a robust cross-validation strategy: 22 cases (20%) held-out for testing, with remaining data 86 cases used for 5-fold cross-validation (stratified by IPH presence). The five resulting models independently generated IPH probability scores, and ensembled predictions calculated by three methods to enhance generalizability: Majority and weighted voting based on binary IPH class and soft voting based on IPH probability scores. Results: Of 418 patients with histological analysis, 130 had CTA studies available. After excluding 22 cases lacking histological consensus, 108 cases were analyzed. In the 22 test cases the mean age was 68.1±8.5 years and 59% were male; In the cross-validation folds age ranged from 67.1±6.9 to 75.3±7.6 years 56-82% were male. Individual models' performance on the test set showed mean ROC-AUC of 0.79 (95% CI: ±0.05) and average precision of 0.85 (±0.05) (Fig 2). The ensemble approach achieved superior performance with soft voting demonstrated 82% ROC-AUC, 77% accuracy, 79% precision, 85% sensitivity, and 82% F1-score for IPH detection, outperforming majority voting and weighted voting (Fig 3). Conclusion: Automated IPH detection in widely available CTA imaging can turn it to a powerful stroke risk assessment tool, supporting personalized prevention strategies and surgical decision-making, ultimately reducing cerebrovascular morbidity and mortality.

Abstract The advances in large language models (LLMs) have demonstrated sophisticated potential for automating complex tasks within the radiology workflow. From radiology report generation and report summarization to data collection for research trials, these models have proven to be powerful tools. However, optimal implementation of these models requires careful adaptation to the specialized medical domain. In addition, these models tend to generate information that is not truthful or factual, which can adversely affect patient care and clinical decisions. Strategies such as fine-tuning and prompt optimization have been shown to be impactful in eliminating these errors. Although these models undergo rapid updates and improvements, understanding the principles of prompt engineering and fine-tuning provides a foundation for evaluating and maintaining the performance of any LLM deployment. The current article aims to review the recent advancements in radiology using fine-tuning and prompt optimization to leverage LLMs’ capabilities. It delves into various techniques within each strategy, their advantages and limitations, and presents a framework to facilitate the practical integration of LLMs into radiology settings.

Abstract Background Automated data extraction from echocardiography reports could facilitate large-scale registry creation and clinical surveillance of valvular heart diseases (VHD). We evaluated the performance of open-source large language models (LLMs) guided by prompt instructions and chain of thought (CoT) for this task. Methods From consecutive transthoracic echocardiographies performed in our center, we utilized 200 random reports from 2019 for prompt optimization and 1000 from 2023 for evaluation. Five instruction-tuned LLMs (Qwen2.0-72B, Llama3.0-70B, Mixtral8-46.7B, Llama3.0-8B, and Phi3.0-3.8B) were guided by prompt instructions with and without CoT to classify prosthetic valve presence and VHD severity. Performance was evaluated using classification metrics against expert-labeled ground truth. Mean squared error (MSE) was also calculated for predicted severity’s deviation from actual severity. Results With CoT prompting, Llama3.0-70B and Qwen2.0 achieved the highest performance (accuracy: 99.1% and 98.9% for VHD severity; 100% and 99.9% for prosthetic valve; MSE: 0.02 and 0.05, respectively). Smaller models showed lower accuracy for VHD severity (54.1%-85.9%) but maintained high accuracy for prosthetic valve detection (>96%). Chain of thought reasoning yielded higher accuracy for larger models while increasing processing time from 2-25 to 67-154 seconds per report. Based on CoT reasonings, the wrong predictions were mainly due to model outputs being influenced by irrelevant information in the text or failure to follow the prompt instructions. Conclusions Our study demonstrates the near-perfect performance of open-source LLMs for automated echocardiography report interpretation with the purpose of registry formation and disease surveillance. While larger models achieved exceptional accuracy through prompt optimization, practical implementation requires balancing performance with computational efficiency.

Background: Previous studies have linked pericoronary artery adipose tissue attenuation to major adverse cardiac event (MACE). However, the association between pericarotid artery fat and prognosis remains underexplored. This study aimed to investigate the association between preoperative pericarotid fat characteristics and the subsequent prognosis in patients undergoing carotid endarterectomy (CEA). Methods: This study included patients with preoperative computed tomography angiography (CTA) followed by CEA. Preoperative neck CTA images were obtained and analyzed with novel methods using 3D Slicer software. We segmented the common and internal carotid arteries. The segment was horizontally trimmed, isolating the region around the bifurcation, extending 15 mm above and 10 mm below. Pericarotid fat was defined as adipose tissue within a 5 mm radial distance from the outer vessel wall, with Hounsfield units (HU) between -190 and -30 (Figure 1, carotid segmentation [green] and pericarotid fat [yellow]) . For each patient, we calculated the HU average and the volume (mm3) of the pericarotid fat. MACE was defined as a composite of all-cause mortality, myocardial infarction, and cerebrovascular events. The relationship between both pericarotid fat characteristics and MACE was investigated. Results: Of the 43 patients initially included, one was excluded due to an inability to perform segmentation caused by chronic total occlusion, leaving 42 patients (age 73 [67-78] years, 33 [76.7%] male). The comparison of mean pericarotid fat HU between the ipsilateral and contralateral sides showed no significant difference (-60.3±11.0 HU vs. -60.9 ±11.3 HU, p=0.437). However, the median ipsilateral fat volume was significantly higher than the contralateral volume (528 [248, 1100] mm3 vs. 328 [184, 697] mm3, p=0.020). When comparing patients with and without MACE, those with MACE had a significantly lower ipsilateral fat density (-59.4±10.2 HU vs. -66.8 ±11.1 HU, p=0.022) and a higher fat volume (468 [233, 746] mm3 vs. 845 [458, 1451] mm3, p=0.044). Kaplan-Meier analysis revealed that higher volume and lower fat density than the median were significant predictors of MACE (log-rank p=0.015 and p=0.042, respectively) (Figure 2). Conclusions: Assessment of pericarotid fat using a novel CTA analytic method in patients undergoing CEA identified ipsilateral lower fat density and higher fat volume as predictors of adverse outcomes.

Background: Previous research has suggested pericarotid fat density to be significantly linked to atherosclerotic plaque components and cerebrovascular events. Inflammatory markers within the plaques are critical in the pathogenesis and progression of carotid artery disease. However, the association between pericarotid fat and inflammatory processes within atherosclerotic plaques remains underexplored. This study hypothesizes that pericarotid fat density is directly associated with the levels of inflammatory markers within atherosclerotic plaques. Methods: Carotid plaques were prospectively collected from patients with carotid artery disease undergoing carotid endarterectomy (CEA). The levels of inflammatory plaque cytokines (MCP-1, IL-8, IFN-γ, IL-10, IL-1β, IL-6, PDGF-AA, PDGF-AB/BB, TNF-α, VEGF) were measured in tissue homogenates using a customized 10-plex Luminex assay. Baseline demographics and clinical data of our population were extracted from electronic health records. We analyzed preoperative neck computed tomography angiography (CTA) images placing regions-of-interest (ROI) in the pericarotid fat on the slice showing maximal stenosis. For each patient, we calculated the average of Hounsfield Units (HU) within the ROI. Spearman correlation test was used to assess the correlation between variables. Results: Among the 43 patients included, the median age was 73 (67, 78) years, and 33 (76.7%) patients were male. Plaque levels of PDGF-AA and VEGF showed significant positive correlations with HU (r=0.330, p=0.030; r=0.380, p=0.011, respectively), as illustrated in Figure 1 . Although plaque levels of IFN-γ demonstrated a significant negative correlation with HU (r=-0.390, p=0.010), the most cases exhibited values close to zero. Conclusions: The assessment of pericarotid fat density on CTA may allow for the estimation of plaque inflammation, suggesting that more fat density is associated with increased inflammatory activity within atherosclerotic plaques.

Background: The bovine aortic arch is a vascular variant in which the left common carotid artery originates from the brachiocephalic trunk rather than directly from the aortic arch. The bovine anatomical variation may alter the blood flow dynamics, potentially contributing to the formation and progression of carotid plaques. In this study, we evaluated the association of the bovine aortic arch and the severity of carotid stenosis. Methods: In this retrospective study, we included patients with carotid artery disease undergoing carotid endarterectomy (CEA) and pre-surgical computed tomography angiography (CTA). CTAs were reviewed to assess the carotid stenosis and the anatomy of the aortic arch. The left carotid artery was classified as normal if it originated from the aortic arch and as bovine if it originated from the brachiocephalic trunk (Figure 1) . Unlimited propensity score matching was performed to balance baseline characteristics between patients with bovine and normal anatomies. The Mann-Whitney U test was applied to compare carotid stenosis between bovine and normal arteries. Results: A total of 203 patients (bovine:23, normal:180) were initially enrolled. Using propensity score matching, 20 patients with bovine anatomy were matched to 45 patients with normal anatomy (Figure 2) . Among the 65 patients included, 47 (74.6%) were male, with a median age of 70 [IQR:65.0-76.0] years. Left-sided carotid artery disease was more prevalent in the bovine compared to normal group (N=12, 66.7% vs. N=16, 36.4%; p=0.030). Additionally, the severity of carotid stenosis in CEA-treated side was higher in patients with bovine arch compared to normal controls (82.5% [76.2-88.7] vs. 70.0% [50.0-95.0]; p=0.030) (Figure 3) . Conclusions: The association of a bovine aortic arch with more severe carotid stenosis suggests that this anatomical variant may be a marker for increased cerebrovascular risk.

Introduction: Calcium scoring computed tomography (CAC) is a widely available and inexpensive modality of screening for atherosclerosis. This study aims to improve the predictive value of CAC by incorporating pericardial fat characteristics in a CatBoost model. Methods: A sample of 7,000 non-contrast clinically indicated CAC scans were randomly selected from a cohort of consecutive Olmsted County MN residents. Known history of atherosclerotic cardiovascular disease or impaired images were excluded. A pericardial fat segmentation model, trained on 191 images of the same dataset with a dice similarity score of 94%, was applied to 6,194 CT scans; Volume and mean Hounsfield unit (mHU) of the output masks were added to age at imaging, radiologist-measured calcium score, male gender, diabetes, and smoking (Figure 1). A 70,20 and 10% split was used to produce the training, validation and test set, respectively. Features were inputted to CatBoost classifier, as the best model tested, to predict four outcomes: acute coronary syndrome (ACS), heart failure (HF), ischemic stroke, and all-cause mortality. Model was evaluated by area under the ROC curve (AUC) on the unseen test set. Results: Mean age was 55.2 (8.8), 55.2 (8.8) and 54.8 (8.5) years in the training, validation and test sets, respectively, and 65%, 67% and 66% of the populations were male. The mean AUC, obtained through bootstrapping with 100 iterations, was 0.81 [0.77-0.85], 0.86 [0.60-0.92], 0.80 [0.73-0.87] and 0.87 [0.81-0.91] for HF, ACS, stroke and death, respectively; The same analysis excluding pericardial fat measurements showed a 3% reduction in the AUC for all of the investigated outcomes (Figure 2). Conclusion: Our study presents a robust risk prediction model for ACS, HF, stroke, and mortality based on CAC scans. The improved performance of the model by incorporating pericardial fat characteristics, highlights the potentials of deep learning in timely inexpensive risk prediction in large populations.

Machine-learning (ML) and deep-learning (DL) algorithms are part of a group of modeling algorithms that grasp the hidden patterns in data based on a training process, enabling them to extract complex information from the input data. In the past decade, these algorithms have been increasingly used for image processing, specifically in the medical domain. Cardiothoracic imaging is one of the early adopters of ML/DL research, and the COVID-19 pandemic resulted in more research focus on the feasibility and applications of ML/DL in cardiothoracic imaging. In this scoping review, we systematically searched available peer-reviewed medical literature on cardiothoracic imaging and quantitatively extracted key data elements in order to get a big picture of how ML/DL have been used in the rapidly evolving cardiothoracic imaging field. During this report, we provide insights on different applications of ML/DL and some nuances pertaining to this specific field of research. Finally, we provide general suggestions on how researchers can make their research more than just a proof-of-concept and move toward clinical adoption.